Taskforce on Women in R

Di Cook

and Heather Turner, Jenny Bryan, Julie Josse, Oliver Keyes, Michael Lawrence,
Kevin O’Brien, Alicia Oshlack, Carolin Strobl

Oct 20, 2016

The R Community - 2007

The R Community - 2007

The R Community - 2015

© Torben Tvedebrink 2015

The R Community - 2015

© Torben Tvedebrink 2015

The R Community - 2015

© Torben Tvedebrink 2015

The R Community - 2016

The R Community - 2016

The R Community - 2016

Where we are now?

• How well are women actually represented in the R community? And in what capacity?
  • Conferences
  • Packages
  • Google Summer of Code
  • R Foundation, R Core
  • Publications
  • R-Ladies

2010 Survey of Packages

• 2010 survey of 1087 maintainers of packages on CRAN/Bioconductor plus contributors on R-forge¹
• 9% of package authors were women (95% CI: 7.1 - 11.0%)
• 32.5% were from the USA

¹Mair, P., Hofmann, E., Gruber, K., Hatzinger, R., Zeleis, A. and Hornik, K. (2015) Motivation, values, and work design as drivers of participation in the R open source project for statistical computing, Proc Nat. Acad. Sci.

Conferences

• useR! has already been working to improve diversity
  • gradual increase in number of female keynotes
  • code of conduct introduced in 2015
  • diversity scholarships & mothers’ room in 2016

useR! 2015

• http://user2015.math.aau.dk
• ~19% participants were women
• ~20% of presentations

R Packages

• All CRAN maintainers (Mar 2016) processed with genderizer
  • uses databases to predict gender from first name
• Supplemented by manual assignment
• 14.8% package authors “female”, 11.4% with P(gender) ≥ 0.8

Other R Project Contributions

• 18% Google Summer of Code 2015 mentors were women (9 individuals)
• In remaining cases, percentages represent one or two women

Publications: JCGS

Where should we be?

US Students

• US National Center for Education Statistics (2013-14)
• Computer science ranges from 18% (Bachelor’s) to 29% (Master’s) female
• Others range from 29% to 58% female

US Occupations

• Data from American Community Survey 2014
• 22% female programmers
• Other scientific/analytic occupations at least 41% female

Goals for R Community

• Developers (CRAN maintainers, GSoC students, ISC funding holders) should at least be comparable to computer science figures
  • > 20% women
• Broader user community (R conference attendees/speakers, RUG members) should at least be comparable to mathematical/natural sciences
  • 30-40% women
  
  

How can we get there?

• Keep data
• Build developer capacity
• Community support

Data Monitoring

Surveys

• Get input from the community on obstacles and ideas for improvements
• Target survey at particular populations, e.g. useR! attendees, local user group attendees, package maintainers, R-help users, StackOverflow users
• Create bank of questions and customise survey for target group
• First survey conducted as part useR! 2016 participants (June 27-30)
• Analysis in the works

Information Central: Website/GitHub

Conferences

• The taskforce made recommendations for useR! 2016
  
  

  Recomendation	Action
  50:50 program committee	7:6 female:male
  50:50 session chairs	7:18
  ≥ 20% women on panels	NA
  put gender stats on website	not done

• Future useR!s
  • maintain current gender balance for invited speakers
  • aim for similar gender balance for tutorials: 28% had female tutor in 2015
• useR! 2017 proposal includes offering childcare

Community building: RLadies

RLadies - London

• Example events
  • drop-in session for beginners
  • tutorials
  • “Tour de R”
  • SF uncubed: recruiting tech event

credit: Chiin-Rui Tan

RLadies - Melbourne Oct 18

RLadies - Melbourne Oct 18

RLadies - Melbourne Oct 18

rOpenSci

• R packages to access scientific data; visualize; document & share data
• Community support: open review process, support with maintenance, package promotion
• Unconference (43% non-male)

credit: Gabriela de Queiroz

Australian rOpenSci 2016

Australian rOpenSci 2016

Community building: Twitter

Recruiting Ambassadors

• WiR emailed prominent women in R community
• Some already active, other prompted to act as ambassadors

I am actually working with the conference committee on funding a female grad student or post doc to go to the conference … I will definitely get the word out
– Mine Çetinkaya-Rundel, email

I’m making myself available to help on giving feedback, helping on ideas, reviews of abstracts, etc. I would love to see more R-ladies presenting at useR!
– Gabriela de Queiroz, message on meetup

R User Groups

• 237 local RUGs
• Heading for 20 RUG meetings per week, somewhere in the world
• Short talks, “show and tell”

RUGs as a Springboard

• Opportunity for women
  • practice public speaking (conference format)
  • get feedback on projects
  • relatively low-cost (short talk, local venue)
• Organisers can help
  • small numbers of speakers: not so hard to balance program
  • try lightning talks, e.g. recent DublinR 2 male, 3 female
  • low risk inviting new people

Female-friendly R Workshops

• Not restricted to R-ladies groups!
  

  

  

  

  
  • female only/ males +1s
  • majority female mentors
• Potential Women in R workshops?
  • beginner workshops in partnership with RUGS
  • summer school for aspiring package developers (Europe 2018)

Google Summer of Code

1. Mentors/students propose 3 month projects
2. Google awards a fixed number of student slots
3. Admins & mentors select projects
   • student earns $5500
   • mentors benefit from contribution to their project

R’s contribution from ideas to implementation managed via GitHub github.com/rstats-gsoc

Google Summer of Code

summerofcode.withgoogle.com

On Ramps for Contributing

• Several R packages on GitHub have a CONTRIBUTING.md
  • often not designed for novices
• Target issues at novices
  • use tags
  • step-by-step instructions

credit: Charlotte Spencer

Mentoring

• Open issue on GitHub and Tweet request for review
• Study group/RUG code swaps
• Consultancy
• Scaling rOpenSci???

R Consortium ISC

• The R Consortium is a framework for the support of R by industry
• Its Infrastructure Steering Committee offers funding for projects
  • software development
  • developing teaching materials
  • documenting best practices
  • standardising APIs
  • research

Funded ISC Projects

• R-Hub: pre-test CRAN packages on multiple platforms, automate package building
• A unified framework for distributed computing in R
• Improving database interface (DBI)
• R Implementation, Optimization and Tooling Workshops (RIOT)
• R Localization (RL10N): enable packages to include translations
• SatRdays: regional conferences
• Simple Features access for R: utilities for geospatial data
• Software Carpentry R instructor training

Take-homes

For Women Starting Out

• Find support: R-ladies/RUG/Study group
• Start small
  • Contribute to existing package
  • Make a package of personal utility functions

• Put on your flame-proof suit and ask stupid questions

  If nm is double, very likely INTEGER(nm)[0] is negative and your C code does nothing at all. Hence it could be fast but useless.
  – Reply to Heather on R-help, 2005

• Publish code to get feedback and credit

For Established Women

• Stay visible
  • Speak at conferences
  • Don’t hang out in clique at meetings
  • Be active online: Twitter, Blog, StackOverflow, GitHub, …
• Mentor other women
  • R-ladies/Female-friendly workshop
  • Encourage contributions on own projects, GSoC
  • Suggest names for conferences, and insist on balance
• Consider applying for ISC funding

For Everyone

• Be aware of (other) women’s contributions
  • Attend talks - even if you can only drop in
  • Invite women to low-key events (RUG, departmental seminar)
• Encourage women’s participation
  • Nominate for speakers, committee members, awards, …
  • Invite onto your project
• Promote welcoming culture
  • Use gender neutral language
  • Give constructive feedback

R Foundation Women in R Taskforce

to improve the participation and experience of women in the R community

	Jenny Bryan CA		Alicia Oshlack AU		Oliver Keyes US
	Di Cook AU		Carolin Strobl CH		Kevin O’Brien IE
	Julie Josse FR		Heather Turner UK		Michael Lawrence US

Summary

• Women are under-represented in both user and developer communities
  • Users ~ 20% women; goal 30-40% women
  • Developers ~ 10% women; goal 20% women
• Women in R Taskforce has been established to address this
  • Current focus: data gathering, communications
  • Future work: outreach